Shared memory for AI agents

Second Brain
One memory for every agent you work with.

A shared, versioned memory that any connected agent can fill, read and keep current. Brief it once. Every agent picks it up.

The handoff problem

Your agents are brilliant.
The handoff is broken.

Old way

You brief every chat again

Your brand lives in one chat. The next agent never got it, so you paste the brief again. And again.

New way

Brief it once

One agent stores what it knows in the brain. Every agent you connect picks it up from there.

How it works

From an empty brain to on-brand work,
in five steps.

Features

A memory
made for AI agents.

Any agent can fill it

Claude, ChatGPT, Codex, Claude Code, Cursor or any connected agent reads and writes the same brain over MCP.

Browse it as a graph

Every concept is a node and every link is a line. Select one to see its description, its sources and what cites it.

Plain Markdown underneath

Each memory is a Markdown file in Open Knowledge Format. It stays readable without a database or a model provider.

Sourced, not summarized

Agents store full documents, not summaries. Each concept lists its sources and who generated it.

Permissions are explicit

Agents authenticate as agents. You decide who can read, write, modify or publish.

Every version can be restored

Each change is saved in History under the name of the agent or person who made it.

Good first brains

Start with what you
keep explaining.

01Brand

Brand and messaging

Naming, voice, colors, type and approved lines, so every agent writes and designs on brand.

02Design

Design system

Tokens, components and patterns, so agents prototype in the real look.

03Engineering

Engineering standards

Review rules, architecture decisions and runbooks your coding agents should follow.

04Project

Project memory

The brief, research, decisions and outcomes for one launch, kept in one place.

Why Second Brain

Open format. Any agent. Your rules.

01

Open format

Every memory is a Markdown file with history. Export it any time.

02

Any agent

Claude, ChatGPT, Codex, Cursor or any connected agent can read and write it.

03

Your rules

You choose who can read, write, modify or publish. Every version can be restored.

FAQ

Questions about
Second Brain

Still curious? Read the Knowledge Base docs, see how to connect your agent, or write to us.

It is a knowledge artifact: a small repository of linked Markdown notes that your AI agents write and read. People browse the same notes as a graph in the app.

One human · Many agents · One memory

Second Brain

by
AgentGrid.io

Give every agent the same working memory.